ArticlePDF Available

Estimating the extinction date of the thylacine with mixed certainty data


Abstract and Figures

The thylacine (Thylacinus cynocephalus), one of Australia's most characteristic megafauna, was the largest marsupial carnivore until hunting, and potentially disease, drove it to extinction in 1936. Though thylacines were restricted to Tasmania for two millennia prior to their extinction, recent “plausible” sightings on the Cape York Peninsula in northern Queensland have emerged, leading some to speculate the species may persist, undetected. Here we show that the continued survival of the thylacine is entirely implausible based on most current mathematical theories of extinction. We present a dataset including physical evidence, expert-validated sightings, and unconfirmed sightings leading up to the present day, and use a range of extinction models, focusing on a Bayesian approach that incorporates all three types of data by modelling valid and invalid sightings as independent processes, to evaluate the likelihood of the thylacine's persistence. Although the last captive individual died in September 1936, our analyses suggest the most likely extinction date would be 1940; other extinction models estimated the thylacine's extinction date between 1936 and 1943, and even the most optimistic scenario suggests the species did not persist beyond 1956. The search for the thylacine, much like similar efforts to “rediscover” other recently extinct charismatic taxa, is likely to be fruitless, especially given that persistence on Tasmania would have been no guarantee the species could reappear in regions that had been unoccupied for millennia. The search for the thylacine may become a rallying point for conservation and wildlife biology, and could indirectly help fund and support critical research in understudied areas like Cape York. However, our results suggest that attempts to rediscover the thylacine will likely be unsuccessful. This article is protected by copyright. All rights reserved
No caption available
No caption available
Content may be subject to copyright.
Research Note
Estimating the extinction date of the thylacine
with mixed certainty data
Colin J. Carlson ,1Alexander L. Bond,2and Kevin R. Burgio 3
1Department of Environmental Science, Policy, and Management, University of California, Berkeley, 130 Mulford Hall, Berkeley, CA
94720, U.S.A.
2Ardenna Research, Potton, Sandy, Bedfordshire SG19 2QA, U.K.
3Department of Ecology and Evolutionary Biology, University of Connecticut, 75 N. Eagleville Road, U-3043, Storrs, CT 06269, U.S.A.
Abstract: The thylacine (Thylacinus cynocephalus), one of Australia’s most characteristic megafauna, was the
largest marsupial carnivore until hunting, and potentially disease, drove it to extinction in 1936. Although
thylacines were restricted to Tasmania for 2 millennia prior to their extinction, recent so-called plausible
sightings on the Cape York Peninsula in northern Queensland have emerged, leading some to speculate the
species may have persisted undetected. We compiled a data set that included physical evidence, expert-validated
sightings, and unconfirmed sightings up to the present day and implemented a range of extinction models
(focusing on a Bayesian approach that incorporates all 3 types of data by modeling valid and invalid sightings
as independent processes) to evaluate the likelihood of the thylacine’s persistence. Although the last captive
individual died in September 1936, our results suggested that the most likely extinction date would be 1940.
Our other extinction models estimated the thylacine’s extinction date between 1936 and 1943, and the most
optimistic scenario indicated that the species did not persist beyond 1956. The search for the thylacine, much
like similar efforts to rediscover other recently extinct charismatic taxa, is likely to be fruitless, especially given
that persistence on Tasmania would have been no guarantee the species could reappear in regions that had
been unoccupied for millennia. The search for the thylacine may become a rallying point for conservation and
wildlife biology and could indirectly help fund and support critical research in understudied areas such as
Cape York. However, our results suggest that attempts to rediscover the thylacine will be unsuccessful and that
the continued survival of the thylacine is entirely implausible based on most current mathematical theories
of extinction.
Keywords: sighting record, Tasmania, Tasmanian tiger
on de la Fecha de Extinci´
on del Tilacino con Datos Mixtos de Certidumbre
Resumen: El tilacino (Thylacinus cynocephalus), una de las especies de megafauna m´
as caracter´
ısticas de
Australia, era el carn´
ıvoro marsupial m´
as grande hasta que la caza, y potencialmente las enfermedades, lo
o a la extinci´
on en 1936. Aunque los tilacinos estuvieron restringidos a Tasmania durante dos milenios
previos a su extinci´
on, recientemente han emergido presuntos avistamientos plausibles en la pen´
ınsula de
Cape York al norte de Queensland, lo que ha llevado a algunos a especular que la especie pudo haber persistido
sin ser detectada. Recopilamos un conjunto de datos que incluy´
o evidencia f´
ısica, avistamientos validados por
expertos, y avistamientos sin confirmaci´
on hasta el d´
ıa de hoy, e implementamos una gama de modelos de
on (enfocados en la estrategia bayesiana que incorpora los tres tipos de datos al modelar avistamientos
alidos e inv´
alidos como procesos independientes) para evaluar la probabilidad de la persistencia del tilacino.
Aunque el ´
ultimo individuo cautivo muri´
o en septiembre de 1936, nuestros resultados sugirieron que la
fecha m´
as probable de extinci´
on habr´
ıa sido en 1940. Nuestros otros modelos de extinci´
on estimaron la
fecha de extinci´
on del tilacino entre 1936 y 1943, y el escenario m´
as optimista indic´
o que la especie no
as all´
a de 1956. Es probable que la b´
usqueda del tilacino, como muchos esfuerzos similares para
Article impact statement: The search for thylacine in Australia is likely to be unsuccessful, and search costs will drain limited conservation
Paper submitted May 7, 2017; revised manuscript accepted October 11, 2017.
Conservation Biology, Volume 00, No. 0, 1–7
2017 Society for Conservation Biology
DOI: 10.1111/cobi.13037
2The Extinct Thylacine
redescubrir a otros taxones carism´
aticos recientemente extintos, sea infruct´
ıfera, especialmente debido a que
la persistencia en Tasmania no habr´
ıa sido garant´
ıa de que la especie pudiera reaparecer en regiones que no
ıan sido ocupadas durante milenios. La b´
usqueda del tilacino podr´
ıa convertirse en un punto de reuni´
para la biolog´
ıa de la conservaci´
on y de la vida silvestre y podr´
ıa ayudar indirectamente a financiar y a
apoyar investigaciones cr´
ıticas en ´
areas que no han sido estudiadas suficientemente, como Cape York. Sin
embargo, nuestros resultados sugieren que los intentos por redescubrir al tilacino no ser´
an exitosos y que la
supervivencia continuada del tilacino es completamente implausible con base en las teor´
ıas matem´
aticas de
on m´
as recientes.
Palabras Clave: registro de avistamientos, Tasmania, tigre de Tasmania
(Thylacinus cynocephalus)
The history of conservation biology has included a few
exceptional errors, in which experts have pronounced
a species extinct only for it to be later rediscovered.
Perhaps, most famous are Lazarus taxa known originally
from the fossil record (e.g., the coelacanth [Latimeria
sp.] and dawn redwood [Metasequoia sp.]), but even
recently declared species extinctions can also sometimes
be overturned. Hope of rediscovering a supposedly ex-
tinct species can inspire volumes of peer-reviewed re-
search, and sometimes a single controversial sighting
(e.g., Fitzpatrick et al. 2005) can be enough to reignite
controversy and justify seemingly endless field inves-
tigation, as in the ongoing search for the Ivory-Billed
Woodpecker (Campephilus principalis) despite all odds
(National Audubon Society 2016). Similarly, in Queens-
land, Australia, 2 unconfirmed sightings in early 2017
have inspired a new search for the thylacine (Thylacinus
The thylacine, or Tasmanian tiger, has been presumed
extinct since the last captive specimen died on 7 Septem-
ber 1936 (Sleightholme & Campbell 2016). Thylacines are
believed to have gone extinct on the Australian mainland
2 millennia ago, thereafter persisting only as Tasmanian
endemics (Paddle 2002). State-sponsored eradication in
Tasmania between 1886 and 1909 caused a devastat-
ing population crash (Sleightholme & Campbell 2016).
This eradication campaign, combined with prey declines,
could have been sufficient extinction pressure (Prowse
et al. 2013), but other research strongly suggests that a
disease similar to canine distemper could have helped
drive the species to extinction (De Castro & Bolker 2005;
Paddle 2012). Although the mechanism has been a topic
of debate, the extinction status of the thylacine has been
essentially unchallenged in peer-reviewed literature. De-
spite this, sightings have continued throughout Tasma-
nia and mainland Australia, often gathering national and
international media attention. In January 2017, 2 uncon-
firmed “detailed and plausible” sightings in the Cape York
Peninsula in northern Queensland sparked renewed in-
terest in the thylacine’s persistence, particularly in the
Australian mainland. Researchers currently intend to in-
vestigate those sightings with camera traps later this year
(James Cook University 2017).
Is there empirical support for this most recent search?
Extinction-date (τE) estimators have been a key part of
parallel debates about the Ivory-billed Woodpecker. What
little work has been done on the thylacine places τEfrom
1933 to 1935; only 1 model (using temporally subset-
ted data) suggests the species might be extant (Fisher
& Blomberg 2012). A subsequent study suggested that
based on search effort, thylacine’s body size, and former
density, they would have been rediscovered by 1983 if
they were still extant (Lee et al. 2017b). These methods
exclude sightings data, but recently developed Bayesian
models differentiate between the processes of verified
and unverified sightings explicitly, allowing researchers
to include uncertain sightings in models as a separate
class of data (Solow & Beet 2014). We applied those
Conservation Biology
Volume 00, No. 0, 2018
Carlson et al. 3
models (and several other extinction date estimators) to
thylacine sightings and asked: What is the probability that
the species might be rediscovered?
Most of the sightings in our data set are from Sleightholme
and Campbell’s (2016) appendix (covering 1937–1980),
which includes 1167 post-1900 sightings classified as a
capture, kill, or sighting and Smith et al.’s (1981) sum-
mary of 243 sightings from 1936 to 1980. Additional
sightings were compiled from Heberle (2004), records
on public websites maintained by interested citizen
groups (, www.thylacineresea, and,
and online news stories from 2007 to 2016. We scored
records as 1, physical evidence (e.g., from bounty
records, museum specimens, or confirmed captures); 2,
confirmed sighting (sightings agreed as valid by experts);
or 3, unconfirmed sighting (sightings not considered valid
by experts) (Fig. 1). For each year from 1900 to 1939,
we used the sighting of the highest evidentiary qual-
ity; captures or killed individuals were physical evidence
(n=101) (Supporting Information). Our assembled data
set spanned from 1900 to 2016 and included the last
confirmed specimen collected in 1937. Thirty-six of the
years included confirmed sightings. There was only 1
instance of an expert sighting without physical evidence
(1932). The remaining sightings in the data set were
unconfirmed sightings. Although this reduction to 1
record per year was required by only some models, we
aimed for data consistency across methods. Because there
were also likely unreported unconfirmed sightings, we
also ran models based on the assumption that an uncon-
firmed sighting occurred annually from 1938 (the 1st year
with no verified sightings or specimens) to 2016, which
produced a marginally higher chance of persistence but
without changing the overall conclusions (Supporting
For all analyses, we considered the species across its
historical range (i.e., mainland Australia and Tasmania)
and included valid sightings from Tasmania alongside
highly questionable sightings from mainland Australia,
despite the species’ supposed extirpation 2 millennia
earlier on the continent. We considered this the only op-
timistic modeling scenario for the thylacine’s persistence
in which recent high-profile sightings could be valid,
even if it represents one of the most biologically implau-
sible scenarios. In Supporting Information, we present an
analysis in which we used only confirmed sightings from
Tasmania, which could be considered a more realistic
analysis of the probability that the thylacine persisted on
Tasmania alone (although this would fail to explain the
most controversial recent sightings throughout mainland
Australia) (Supporting Information).
Bayesian Extinction Estimators
Methods to estimate extinction dates from time series
data have been popular in conservation biology since
the 1990s, but the majority fail to account for the vari-
ability in quality and certainty of most sighting records
(Boakes et al. 2015). However, several Bayesian methods
have been developed recently that incorporate variable
sighting quality, including unconfirmed sightings, in es-
timation of extinction date. These methods rely on the
assumption that the probability a species is extinct (an
event E), based on a time series of sightings t=(t1,...,tn),
is expressed by Bayes’ theorem:
E)(1 p(E)),(1)
Figure 1. Thylacine sighting
data (confirmed specimens,
absolute and certain form
of evidence; confirmed
sightings, expert-verified
sightings of intermediate
level of certainty;
unconfirmed sightings,
controversial sightings or
indirect evidence based on
scat or tracks, weakest
source of evidence).
Conservation Biology
Volume 00, No. 0, 2018
4The Extinct Thylacine
and where ¯
Eis the scenario that the species is not extinct
within the period in question. The prior probability of
extinction P(E) can often be hard to define, although
it is sometimes uninformatively set to 0.5 for explicit
estimation of P(E|t). However, the Bayes factor can be
used as a test of support for Ewhere
(although it can be formatted in reverse in some studies,
as in the case of Lee et al. [2014], given the probability
that a species persists). The relationship between the
Bayes factor and the probability of the species’ extinction
is given as
Consequently, with an uninformative prior,
and given a sufficiently large Bayes factor, the probability
of persistence is
A handful of extinction-date estimation methods have
been developed using Bayesian frameworks that allow es-
timation of the Bayes factor and thereby support hypoth-
esis testing. The set of models on which we focused were
first developed by Solow et al. (2012), who proposed a
method in which all sightings leading up to a date tL(the
date of the last certain sighting) were certain and those
after tLwere uncertain. Valid and invalid sightings were
generated by stationary Poisson processes with different
rates, but certain and uncertain sightings had the same
rate (Solow et al. 2012). A more recent revision (Solow
& Beet 2014) proposes 2 major modifications. In the first
modification (model 1), the same assumptions are made
as in the original method, except that uncertain sight-
ings are permitted before tL.The second modification
(model 2, which we used) differs more notably in that it
also treats certain and uncertain sightings as independent
Poisson processes. This model is recommended for cases
in which certain and uncertain sightings “differ qualita-
tively,” as in our study. For example, we note that blurry
photographic or video evidence and crowdsourced sight-
ing records from citizen groups are unique issues for later,
uncertain thylacine sightings. Therefore, this model is
more appropriate than model 1 (or the model from Solow
and Beet 2012) for our study.
In Solow and Beet’s (2014) model 2, although the rate
of valid sightings is likely to change leading up to an ex-
tinction event, after extinction that rate remains constant
(at zero) and all sightings are presumed unconfirmed. The
sighting data set toccurs over an interval [0,T), where 0
τE<T. During the interval [0, τE), valid sightings occur at
rate , whereas invalid sightings occur at rate , meaning
that valid sightings occur at proportion
Confirmed sightings occur, at an independently deter-
mined rate, which divides the data set of sightings tinto
confirmed sightings tcand unconfirmed sightings tu.The
likelihood of the data conditional on τEis given as
These 2 values are calculated using nc(the number of
confirmed sightings, all before τE)andnu(the number
of unconfirmed sightings), where nu(τE) are the subset
recorded before τE,andωacts as a dummy variable re-
placing :
Likelihood p(t|τE) is calculated as the product of those
The posterior probability that the species became ex-
tinct in the interval (0,T), which we denoted as an event
E(with alternate hypothesis ¯
E), is given for a prior
The alternate probability p(t|¯
E) can be calculated by
evaluating the same expression given above for p(t|τE)
at τE=T.The Bayes factor is given as
and expresses the relative support for the hypothesis
that extinction happened in the interval [0,T). The most
subjectivity in the method is therefore introduced in se-
lecting the prior τE. Solow and Beet (2014) suggest 3
possibilities: a linear or exponential decline after the last
confirmed sighting or a uniform (uninformative) prior.
We elected to use the uniform prior in all the models
because it makes the least constrained assumption about
the species’ likely extinction status.
In addition to the models developed by Solow and
Beet (2014), we also included another Bayesian model
(Lee et al. 2014) that builds on similar foundational
work (Solow 1993; Solow et al. 2012). Like Solow and
Beet’s (2014) model 2, the model from Lee et al. (2014)
treats confirmed and unconfirmed sightings as separate
processes, and Lee et al. (2014) make slightly different
recommendations regarding how to select a prior
Conservation Biology
Volume 00, No. 0, 2018
Carlson et al. 5
probability that a given sighting is valid, but the overall
intention of the model is largely the same. The approach
in Lee et al. (2014) is also implemented stochastically
with an Monte Carlo Markov Chain (MCMC) approach in
BUGS, whereas Solow and Beet’s (2014) model calculates
likelihoods explicitly. We implemented the model from
Lee et al. (2014) with some of the simplest possible
assumptions: the false positive rate for confirmed sight-
ings is 0, whereas the false positive rate for unconfirmed
sightings samples from a large uniform distribution.
That method can also be implemented more flexibly by
assigning different priors to different categories of evi-
dence, as Lee et al. (2014) suggest. However, rather than
use somewhat arbitrarily chosen priors to differentiate
among our uncertain reports (a refinement with limited
benefits, per a recent study [Lee et al. 2017a]), we
simply divided our data into confirmed and unconfirmed
sightings. Other Bayesian models existed, but we chose
to include the 2 appropriate recently developed Bayesian
methods with available code (Boakes et al. 2015).
Other Extinction Estimators
For the sake of completeness, we also included several
other widely used non-Bayesian estimators with varying
levels of complexity (Rivadeneira et al. 2009; Boakes et al.
2015) that could be readily calculated with the R package
sExtinct (Clements 2013a). Were we to include every
unconfirmed, controversial sighting up to 2016 and to
treat these as valid sightings (because these models make
no such distinction), all methods would indicate that the
species is unequivocally extant. Consequently, we lim-
ited the implementation of other methods to 2 practical
applications and examined how results changed by either
including only confirmed, uncontroversial specimens or
both confirmed specimens and confirmed sightings (Sup-
porting Information).
Among the methods we included, Robson and Whit-
lock (1964) suggested a nonparametric method based on
the last 2 sightings (with an associated Pvalue):
This produced the latest thylacine τE(Supporting In-
formation), as expected, given that the method can be
prone to severe overestimation. Burgman et al. (1995)
used the length of the period of observation, the num-
ber of years with and without records, and the length
of the longest consecutive set of years with records to
derive a combinatoric probability of unobserved pres-
ence. Similarly, Strauss and Sadler (1989) developed a
Bayesian method focused on the discrepancy between
the observed interval of sightings (between the first and
last sighting) and the true range of a species in the fossil
record. Setting a precedent on which more current meth-
ods are based, Solow (1993) in his original method as-
sumes that sightings are a stationary Poisson process, in
which the probability of persistence is
However, a subsequent formulation makes the more
accurate assumption that sightings follow a truncated ex-
ponential distribution, declining until extinction (Solow
2005). Finally, the optimal linear estimator (OLE) method
(Roberts & Solow 2003) is the most robust nonparametric
extinction estimator (Clements et al. 2013) and is based
on a subset of the last ssightings of ktotal,
where bis a vector of s1s, and is a square matrix of
dimension swith typical element
ij =(2ˆυ+i)υ+j)
ln tktks+1
Based on model 2 from Solow and Beet (2014), the most
likely value for τEwas 1940; the posterior likelihood
declined rapidly thereafter (Fig. 2). The probability the
thylacine is extant was extremely low (Bayes factor =
6.08912 ×1013; equivalently, an odds ratio of 1 in 60.9
trillion). Using Lee et al.’s (2014) method, the probability
of persistence was estimated to be zero by 1940. All of
the non-Bayesian estimating models agreed with these
findings. Using only confirmed specimens provided an
OLE estimated extinction date of 1938 (95% CI 1937–
1943). Adding confirmed sightings did not change the
estimated extinction date or confidence interval. Robson
and Whitlock’s (1964) method gave τEas 1956, the latest
estimate (Table 1 & Supporting Information).
Based on the results of our Solow and Beet’s (2014) model
2, it remains plausible that the thylacine’s extinction
could have occurred up to a decade later than believed.
But for thylacines to appear in 2017, especially where
they are believed to have been absent for 2 millennia,
is highly implausible. The 2 sightings from Cape York
described as “detailed” and “plausible” (Hunt 2017) may
Conservation Biology
Volume 00, No. 0, 2018
6The Extinct Thylacine
Figure 2. The posterior probability of
a given thylacine extinction date (τE)
scaled by the area under the entire
likelihood curve. In Solow and Beet’s
(2014) model, specimen-based
records are treated separately and as
certain observations; consequently,
evaluation begins in 1937, the year
of the last certain sighting (i.e.,
extinction prior to that date is not
Table 1. Main estimates for thylacine extinction dates.
Explicit inclusion
of uncertainty? τE
Roberts and Solow (2003) no 1938
Solow and Beet (2014)byes 1940
Lee et al. (2014) yes 1940
Strauss and Sadler (1989) no 1940
Solow (1993) no 1941
Solow (2005) no 1942
Burgman et al. (1995) no 1945
Robson and Whitlock (1964) no 1956
aParametric estimates, except the optimal linear estimator, are cal-
culated with a cutoff of α<0.05.
bGiven by the year with the highest posterior likelihood.
cGiven by the posterior probability reaching zero.
be so from a strictly zoological perspective, but from a
modeling standpoint, they fit neatly into a pattern of on-
going, false sightings that follow nearly any high-profile
extinction. However, models can be wrong, and new
data may overturn a century of common knowledge in
what could be one of the most surprising rediscoveries
in conservation history.
The hope of rediscovering extinct species is one of
the most powerful emotional forces in conservation and
can bring attention to threatened species and ecosystems
while igniting public interest (and funding) (Clements
2013b). The search for the thylacine may reap those ben-
efits, and the proposed 2017 search has already gathered
significant attention from journalists and on social me-
dia. Moreover, the data that will be collected during the
search for the thylacine in Cape York may be invaluable
for other conservation studies. But the ongoing search for
extinct species, in the broader sense, likely diverts critical
funds required desperately for the conservation of nearly
extinct species. About 7% of some invertebrate groups
may already have gone extinct, at which rate 98% of all
extinctions would be entirely undetected (R´
egnier et al.
2015). Globally, 36% of mammal species are threatened
with extinction (classified as vulnerable, endangered of
critically endangered by the International Union for Con-
servation of Nature), including 27% of native Australian
mammals (IUCN 2016), and limited resources can be bet-
ter spent reversing those declines than chasing the ghosts
of extinction past.
We thank A. Beet for the original Matlab code used in
Solow and Beet (2014), A. Butler (Biomathematics and
Statistics Scotland) for translating the Matlab code into
R, and L. Bartlett for helpful criticism and feedback. We
thank 2 anonymous reviewers for helpful feedback, and
particularly for providing useful code adapting the Open-
Bugs scripts of Lee et al. (2014) for our study.
Supporting Information
The results of supplemental extinction models (Appen-
dices S1–S3) and our data set of thylacine sightings
(Appendix S4) are available online. The authors are solely
responsible for the content and functionality of these
materials. Queries (other than absence of the material)
should be directed to the corresponding author.
Supporting lnformation
Literature Cited
Boakes EH, Rout TM, Collen B. 2015. Inferring species extinction: the
use of sighting records. Methods in Ecology and Evolution 6:678–
Conservation Biology
Volume 00, No. 0, 2018
Carlson et al. 7
Burgman MA, Grimson RC, Ferson S. 1995. Inferring threat from scien-
tific collections. Conservation Biology 9:923–928.
Clements C. 2013a. sExtinct: calculates the historic date of extinction
given a series of sighting events. R package version 1.1. R Foundation
for Statistical Computing, Vienna.
Clements CF. 2013b. Public interest in the extinction of a species may
lead to an increase in donations to a large conservation charity.
Biodiversity and Conservation 22:2695–2699.
Clements CF, Worsfold NT, Warren PH, Collen B, Clark N, Blackburn
TM, Petchey OL. 2013. Experimentally testing the accuracy of an ex-
tinction estimator: Solow’s optimal linear estimation model. Journal
of Animal Ecology 82:345–354.
De Castro F, Bolker B. 2005. Mechanisms of disease-induced extinction.
Ecology Letters 8:117–126.
Fisher DO, Blomberg SP. 2012. Inferring extinction of mammals from
sighting records, threats, and biological traits. Conservation Biology
Fitzpatrick JW, et al. 2005. Ivory-billed Woodpecker (Campephilus
principalis) persists in continental North America. Science
Heberle G. 2004. Reports of alleged thylacine sightings in Western
Australia. Conservation Science Western Australia 5:1–5.
Hunt E. 2017. ‘Sightings’ of extinct Tasmanian tiger prompt search
in Queensland. The Guardian. Available from https://www.
sighting-search-thylacine-queensland-australia (accessed March
IUCN (International Union for Conservation of Nature). 2016. The IUCN
red list of threatened species. Version 2016-3. IUCN, Gland, Switzer-
land. Available from (accessed March
James Cook University. 2017. Press release: FNQ search for the Tas-
manian tiger. James Cook University, Douglas, Queensland. Avail-
able from
search-for-the-tasmanian-tiger (accessed March 2017).
Lee TE, Bowman C, Roberts DL. 2017a. Are extinction opinions extinct?
PeerJ 5:e3663.
Lee TE, Fisher DO, Blomberg SP, Wintle BA. 2017b.Extinctorstillout
there? Disentangling influences on extinction and rediscovery helps
to clarify the fate of species on the edge. Global Change Biology
Lee TE, McCarthy MA, Wintle BA, Bode M, Roberts DL, Burgman MA.
2014. Inferring extinctions from sighting records of variable relia-
bility. Journal of Applied Ecology 51:251–258.
National Audubon Society. 2016. April 11. The quest for the Ivory-Billed
Woodpecker heads to Cuba. National Audubon Society, New York.
Available from
billed-woodpecker-heads-cuba (accessed March 2017).
Paddle R. 2002. The last Tasmanian tiger: the history and extinction
of the thylacine. Cambridge University Press, Cambridge, United
Paddle R. 2012. The thylacine’s last straw: epidemic disease in
a recent mammalian extinction. Australian Zoologist 36:75–
Prowse TA, Johnson CN, Lacy RC, Bradshaw CJ, Pollak JP, Watts MJ,
Brook BW. 2013. No need for disease: testing extinction hypotheses
for the thylacine using multi-species metamodels. Journal of Animal
Ecology 82:355–364.
egnier C, Achaz G, Lambert A, Cowie RH, Bouchet P, Fontaine B. 2015.
Mass extinction in poorly known taxa. Proceedings of the National
Academy of Sciences 112:7761–7766.
Rivadeneira MM, Hunt G, Roy K. 2009. The use of sighting records
to infer species extinctions: an evaluation of different methods.
Ecology 90:1291–1300.
Roberts DL, Solow AR. 2003. Flightless birds: When did the dodo be-
come extinct? Nature 426:245.
Robson D, Whitlock J. 1964. Estimation of a truncation point.
Biometrika 51:33–39.
Sleightholme SR, Campbell CR. 2016. A retrospective assessment of
20th century thylacine populations. Australian Zoologist 38:102–
Smith SJ, et al. 1981. The Tasmanian tiger, 1980: a report on an investi-
gation of the current status of thylacine Thylacinus cynocephalus,
funded by the World Wildlife Fund Australia. National Parks and
Wildlife Service, Hobart, Tasmania.
Solow A, Smith W, Burgman M, Rout T, Wintle B, Roberts D. 2012. Un-
certain sightings and the extinction of the Ivory-Billed Woodpecker.
Conservation Biology 26:180–184.
Solow AR. 1993. Inferring extinction from sighting data. Ecology
Solow AR. 2005. Inferring extinction from a sighting record. Mathemat-
ical Biosciences 195:47–55.
Solow AR, Beet AR. 2014. On uncertain sightings and inference about
extinction. Conservation Biology 28:1119–1123.
Strauss D, Sadler PM. 1989. Classical confidence intervals and Bayesian
probability estimates for ends of local taxon ranges. Mathematical
Geology 21:411–427.
Conservation Biology
Volume 00, No. 0, 2018
... However, hundreds of unverified observations have been reported in Tasmania over subsequent decades. Carlson et al. (2018) used physical evidence and uncertain sightings to analyze the record of thylacine encounters in Tasmania from 1900 to 2015. They concluded that extinction was likely by 1940 and that there was virtually no chance of persistence to the present day (1 in 112 trillion against). ...
... For these unverified sightings, the quality of reports and associated documentation did not noticeably change preto post-1930s (Rounsevell & Smith 1982), except that no more specimens were confirmed. However, there have also been many reports from mainland Australia of quality comparable to those from Tasmania (Lang 2014;Carlson et al. 2018), which argues for a down weighting in the reliability of eyewitness reports alone. Although technological advances (e.g., automated thermal imaging cameras) have improved the likelihood of detection, they are still not distributed widely across the thylacine's plausible relictual geographic range, and these recent innovations say nothing about the possibility of persistence over the few decades following the last verified physical evidence. ...
Article impact statement:Due to reporting biases, it is premature to conclude that thylacine extinction occurred soon after the last captive specimen died.
... Uncertain records were those that could either not be linked directly to the main island of Tristan (i.e., they could have been referring to Inaccessible, Nightingale or Gough islands), or if detailed ornithological notes were made on two of the three species, but not the third. We follow the template of other studies that use sighting date records with mixed uncertainty (Solow and Beet 2014;Carlson et al. 2018a), and divided presence records into three tiers: confirmed and verifiable (e.g., a museum specimen); confirmed, but unverifiable (e.g., an expert sighting); or unconfirmed, but plausible (e.g., second-hand records, or records of questionable identification). ...
Full-text available
The overwhelming majority of avian extinctions have occurred on islands, where introduced predators, habitat loss, disease, and human persecution have resulted in the loss of over 160 species in the last 500 years. Understanding the timing and causes of these historical extinctions can be beneficial to identifying and preventing contemporary biodiversity loss, as well as understanding the nature of island ecosystems. Tristan da Cunha (henceforth “Tristan”), the most remote inhabited island in the world, has lost three species from the main island since permanent human settlement in 1811—the Tristan Moorhen (Gallinula nesiotis), Inaccessible Finch (Nesospiza acunhae acunhae), and Tristan Albatross (Diomedea dabbenena). We used recently developed Bayesian methods, and sightings of mixed certainty compiled from historical documents, to estimate the extinction date of these three species from Tristan based on specimens. We estimate that all three species were likely extirpated from Tristan between 1869 and 1880 following a period of significant habitat alteration and human overexploitation, and only the albatross had a high probability of persistence when Black Rats (Rattus rattus) arrived in 1882, the previously assumed cause of extinction for all three species. Better estimates of extinction dates are essential for understanding the causes of historical biodiversity loss, and the combination of historical ecology with modern statistical methods has given us novel insights into the timing and therefore the causes of extinctions on one of the most isolated islands in the world.
... In "Estimating the Extinction Date of the Thylacine with Mixed Certainty Data," we (Carlson et al. 2018a) used the sighting record, including controversial post-1936 sightings, to model the probability that the thylacine has been classified accurately as extinct. We found astronomically low odds that the thylacine is extant and argue that a camera-trap search for the species in Cape York, northern Queensland, may be motivated by false hope. ...
Full-text available
Article impact statement: All available models suggest the thylacine remains extinct until rediscovered; hope remains for other species.
Full-text available
Spatially explicit population models (SEPMs) can simulate spatio‐temporal changes in species’ range dynamics in response to variation in climatic and environmental conditions, and anthropogenic activities. When combined with pattern‐oriented modelling methods, ecological processes and drivers of range shifts and extinctions can be identified, and plausible chains of causality revealed. The open‐source multi‐platform R package poems provides functionality for simulating and validating projections of species’ range dynamics using stochastic, lattice‐based population models. Built‐in modules allow parameter uncertainty to propagate through to model simulations, with their effects on species’ range dynamics evaluated using Approximate Bayesian Computation. These validation procedures identify models with the structural complexity and parameterisation needed to simulate the effects of past changes in climate, environment and human activities on species’ range shifts and extinction risk. We illustrate the features and versatility of poems by simulating the historical decline and extinction of the Thylacine (Thylacinus cynocephalus), an icon of recent extinctions in Australia. We show that poems can reveal likely ecological pathways to extinction using pattern‐oriented methods, providing validated projections of the range collapse and population decline of threatened species. By providing flexible and extendable modules for building and validating SEPMs of species’ range dynamics, poems allows the effects of past and future threats on species’ populations to be quantified using well‐parameterised, structurally realistic models, with important generative mechanisms. Since poems can directly unravel ecological processes of species responses to global change, and strengthen predictions of range shifts and extinction risk—within a flexible, R‐based environment—we anticipate that poems will be of significant value to ecologists, conservation managers and biogeographers.
Extinction risk is often associated with intrinsic species traits such as larger size, higher trophic level, narrower habitat niche or smaller distribution area. Despite this, fast extinctions can also occur in species that apparently do not exhibit any of these traits. The Andalusian Buttonquail (Turnix sylvaticus sylvaticus) is a critically endangered taxon, which barely survives in a single population in western Morocco. Here, we describe how this taxon with a formerly wide distribution range, high reproductive rates, low trophic level in the food chain, small size and apparently coarse habitat requirements, is heading towards extinction. By means of environmental niche modelling, we outline its historical distribution and then at a regional scale (Andalucía) we explore the role of historical land use changes and human population trends in the rapid decline of the species. PCA of environmental variables showed that its distribution was mainly determined by low continentality and aridity. Since the nineteenth century, the decline in the extent of occurrence has been above 99.99%. PCA of land use changes showed that areas with a higher probability of historical presence have suffered more intense agricultural intensification and afforestation processes. These areas have also been those which have suffered higher human population pressure and development. Any conservation efforts should focus on maintaining coexistence of the species with humans.
Despite recent interest in invertebrate extinctions, most go undocumented due to chronic data deficiencies. Here, we investigate the absence of the New England medicinal leech (Macrobdella sestertia), a species not sighted in the last decade, but which has been rediscovered several times since it was described in 1886. We collate all known records (19 specimens from nine independent sightings) of this rare species, and use sighting-based extinction date estimators to evaluate the odds that populations of M. sestertia might persist. Using all available data and several different models, we find that while the species may be extinct, there is also not enough evidence to eliminate the hypothesis that the species remains extant. Given that the geographic range of M. sestertia is now known to extend much further south than previously thought, we recommend protections for M. sestertia should be expanded at state and potentially federal levels. We also recommend that systematists should continue to revisit Macrobdella records in natural history collections, which may turn up new evidence of its range, habitat preference, and possible persistence. We conclude by discussing implications of this work for the broader picture of leech conservation in Europe and North America.
Full-text available
Extinction models vary in the information they require, the simplest considering the rate of certain sightings only. More complicated methods include uncertain sightings and allow for variation in the reliability of uncertain sightings. Generally extinction models require expert opinion, either as a prior belief that a species is extinct, or to establish the quality of a sighting record, or both. Is this subjectivity necessary? We present two models to explore whether the individual quality of sightings, judged by experts, is strongly informative of the probability of extinction: the ‘quality breakpoint method’ and the ‘quality as variance method’. For the first method we use the Barbary lion as an exemplar. For the second method we use the Barbary lion, Alaotra grebe, Jamaican petrel and Pohnpei starling as exemplars. The ‘quality breakpoint method’ uses certain and uncertain sighting records, and the quality of uncertain records, to establish whether a change point in the rate of sightings can be established using a simultaneous Bayesian optimisation with a non-informative prior. For the Barbary lion, there is a change in subjective quality of sightings around 1930. Unexpectedly sighting quality increases after this date. This suggests that including quality scores from experts can lead to irregular effects and may not offer reliable results. As an alternative, we use quality as a measure of variance around the sightings, not a change in quality. This leads to predictions with larger standard deviations, however the results remain consistent across any prior belief of extinction. Nonetheless, replacing actual quality scores with random quality scores showed little difference, inferring that the quality scores from experts are superfluous. Therefore, we deem the expensive process of obtaining pooled expert estimates as unnecessary, and even when used we recommend that sighting data should have minimal input from experts in terms of assessing the sighting quality at a fine scale. Rather, sightings should be classed as certain or uncertain, using a framework that is as independent of human bias as possible.
Full-text available
Each year, two or three species that had been considered to be extinct are rediscovered. Uncertainty about whether or not a species is extinct is common, because rare and highly threatened species are difficult to detect. Biological traits such as body size and range size are expected to be associated with extinction. However, these traits, together with the intensity of search effort, might influence the probability of detection and extinction differently. This makes statistical analysis of extinction and rediscovery challenging. Here we use a variant of survival analysis known as cure rate modelling to differentiate factors that influence rediscovery from those that influence extinction. We analyse a global dataset of 99 mammals that have been categorised as extinct or possibly extinct. We estimate the probability that each of these mammals is still extant, and thus estimate the proportion of missing (presumed extinct) mammals that are incorrectly assigned extinction. We find that body mass and population density are predictors of extinction, and body mass and search effort predict rediscovery. In mammals, extinction rate increases with body mass and population density, and these traits act synergistically to greatly elevate extinction rate in large species that also occurred in formerly dense populations. However, when they remain extant, larger-bodied missing species are rediscovered sooner than smaller species. Greater search effort increases the probability of rediscovery in larger species of missing mammals, but has a minimal effect on small species, which take longer to be rediscovered, if extant. By separating the effects of species characteristics on extinction and detection, and using models with the assumption that a proportion of missing species will never be rediscovered, our new approach provides estimates of extinction probability in species with few observation records and scant ecological information.
Full-text available
Since the 1980s, many have suggested we are in the midst of a massive extinction crisis, yet only 799 (0.04%) of the 1.9 million known recent species are recorded as extinct, questioning the reality of the crisis. This low figure is due to the fact that the status of very few invertebrates, which represent the bulk of biodiversity, have been evaluated. Here we show, based on extrapolation from a random sample of land snail species via two independent approaches, that we may already have lost 7% (130,000 extinctions) of the species on Earth. However, this loss is masked by the emphasis on terrestrial vertebrates, the target of most conservation actions. Projections of species extinction rates are controversial because invertebrates are essentially excluded from these scenarios. Invertebrates can and must be assessed if we are to obtain a more realistic picture of the sixth extinction crisis.
Full-text available
A range of mathematical models has been developed to infer whether a species is extinct based on a sighting record. Although observations have variable reliability, current methods for detecting extinction do not differentiate observation qualities. A more suitable approach would consider certain and uncertain sightings throughout the sighting period. We consider a small population system, meaning we assume sighting rates are constant and the population is not declining. Based on such an assumption, we develop a Bayesian method that assumes that certain and uncertain sightings occur independently and at uniform rates. These two types of sightings are connected by a common extinction date. Several rates of false sightings can be calculated to differentiate between observation types. Prior rates of false and true sightings, as well as a prior probability that the species is extant, are included. The model is implemented in OpenBugs, which uses Markov chain Monte Carlo (MCMC). Based on records of variable reliability, we estimate the probability that the following species are extinct: Caribbean seal Monachus tropicalis, grey, black-footed ferret Mustela nigripes, Audubon & Bachman, greater stick-nest rat Leporillus conditor, Sturt, and lesser stick-nest rat Leporillus apicalis, Gould. As further examples, Birdlife International provided the sighting records for the Alaotra grebe Tachybaptus rufolavatus, Delacour, Jamaica petrel Pterodroma caribbaea, Carte, and Pohnpei mountain starling Aplonis pelzelni, Finsch, with prior probabilities for extinction. The results are compared with existing methods, which ignore uncertain sightings. We find that including uncertain sightings can considerably change the probability that the species is extant, in either direction. However, in our examples, including the quality of the uncertain sighting made little difference. When we ignore uncertain sightings, our results agree with existing methods, especially when the last sighting was near the end of the sighting period. Synthesis and applications. Estimating the probability that a species is extinct based on sighting records is important when determining conservation priorities and allocating available resources into management activities. Having a model that allows for certain and uncertain observations throughout the sighting period better accommodates the realities of sighting quality, providing a more reliable basis for decision-making.
Full-text available
The extinction of many species can only be inferred from the record of sightings of individuals. Solow et al. (2012, Uncertain sightings and the extinction of the Ivory-billed Woodpecker. Conservation Biology 26:180-184) describe a Bayesian approach to such inference and apply it to a sighting record of the Ivory-billed Woodpecker (Campephilus principalis). A feature of this sighting record is that all uncertain sightings occurred after the most recent certain sighting. However, this appears to be an artifact. We extended this earlier work in 2 ways. First, we allowed for overlap in time between certain and uncertain sightings. Second, we considered 2 plausible statistical models of a sighting record. In one of these models, certain and uncertain sightings that are valid arise from the same process whereas in the other they arise from independent processes. We applied both models to the case of the Ivory-billed Woodpecker. The result from the first model did not favor extinction, whereas the result for the second model did. This underscores the importance, in applying tests for extinction, of understanding what could be called the natural history of the sighting record.
Full-text available
The ivory-billed woodpecker (Campephilus principalis), long suspected to be extinct, has been rediscovered in the Big Woods region of eastern Arkansas. Visual encounters during 2004 and 2005, and analysis of a video clip from April 2004, confirm the existence of at least one male. Acoustic signatures consistent with Campephilus display drums also have been heard from the region. Extensive efforts to find birds away from the primary encounter site remain unsuccessful, but potential habitat for a thinly distributed source population is vast (over 220,000 hectares).
The thylacine (Tasmanian tiger Thylacinus cyanocephalus) was Australia's largest carnivorous marsupial at the time of arrival of Europeans. The animal was the size and shape of a large dog. Thylacines lived in Tasmania until 1936 when the last one died in captivity at Hobart Zoo. There have been a few hundred sightings in Tasmania since then, but none have been confirmed in a scientific sense. The Tasmanian National Parks and Wildlife Service considers that the thylacine is probably extinct in Tasmania. Fossil remains of thylacines have been discovered in all Australian States and New Guinea but they are considered by scientists to have been extinct on the mainland for some 3000 years. There have been alleged thylacine sightings in all of the mainland States but as in Tasmania, none of the sightings have been confirmed scientifically. This paper provides some data derived from 203 alleged thylacine reports from Western Australia, brought to the attention of the Department of Conservation and Land Management (CALM) and/or the Mystery Animal Research Centre of Australia (MARCA) to 1998.
The thylacine (Thyladnus cynocephalus) once ranged widely across Tasmania from the east to west coasts. The authors present the first comprehensive study to delineate the extent of the thylacine's post-1900 range, based on the retrospective analysis of 1167 geo-referenced capture, kill, and confirmed sighting reports, from 1900 to 1940. They examine the probable causes of population collapse, and discuss the possibility that the species survived into the 1940s and beyond.
1.Accurate measures of extinction are needed in biodiversity monitoring and conservation management but ascertaining the exact time at which a species became extinct is difficult since a small population may go undetected for many years.2.For little-studied species, often the only information available is historical sighting data. Several statistical tests have been developed which use this information to make inferences about a species’ extinction. The increasing array of methods can present a daunting choice.3.We review the more frequently cited methods, for each model explaining its assumptions, the data required, the scenarios it was developed for and power considerations, if known. We provide guidance on selecting the most appropriate method for a particular sighting record.4.We give examples from the literature to show how the methods have been usefully applied across conservation research, informing conservation decision-making and extinction inference.This article is protected by copyright. All rights reserved.
While anecdotal accounts exist in the literature of epidemic disease as a significant factor in recent mammalian extinctions, harder data has not previously been presented. The statistics from the deliberate killing of thylacines as a pest species support contemporary records at the turn of the twentieth century, of an epidemic disease in thylacines and other marsupi-carnivores. For the first time, detailed symptoms and statistics of the disease are presented, as recorded by museum staff, and zoological-garden curators and veterinarians. It is argued that the effects of the disease in captivity, which more than halved thylacine longevity, and preferentially affected juveniles, are conformable with the expression of the disease recorded amongst wild thylacines, and demand a recognition of the importance of this disease as a major factor in the thylacine's recent extinction, and its consideration as an influential factor on the distribution and population dynamics of extant marsupi-carnivores. It also practically demonstrates the obvious potential for disease to have been involved in megafaunal extinctions in the past.